FlowsBook a callSign inCreate free accountSign up
Workflows by industry

How transit agencies find the areas where people live far from a stop

Put the population grid next to your own stop list and you see which blocks have many residents and no stop within a short walk. How planners do it, step by step.

Part of the guide Workflow automation

By the xMap team · · 7 min read

A transit agency finds the parts of a city that are poorly served by laying two maps over each other: where people live, in small cells of about a tenth of a square kilometre, and where its stops are. For every populated cell it measures the distance to the nearest stop, or counts the stops within a short walk, and ranks the cells with the most residents and the fewest stops. That ranked list is the start of a network review, a new route or an on-demand service zone.

This article is for network planners, service designers and the analysts who support them in bus, tram and metro agencies, and for consultancies that do this work for cities. It explains the two measures that matter, why a straight line from a stop is not the same as a walk, a made-up example in Paris, the traps in the data, and an AI employee that redoes the analysis whenever the network or the city changes.

Two questions, not one

Planners usually ask "how many people live within 400 or 500 metres of a stop?" That is a coverage figure for the whole network, and it hides where the gaps are. The more useful question is the other way round: for each block where people live, how far is the nearest stop, and how many people are in that block? A block of 3,000 residents that is 900 metres from any stop matters more than an empty industrial estate the same distance away.

  • Distance to the nearest stop tells you how bad the gap is for the people in a block.
  • Stops within a walk tells you about choice: one stop served once an hour is a weaker offer than four stops on frequent lines.
  • Residents in the block tells you how many people the gap affects, and so where to start.

Neither measure says anything about frequency, hours or where people want to go. Those come from your own timetable and travel surveys. What the location data adds is the base layer: who lives where, and how far they are from the network as it stands.

Where the population figures come from, and where they do not exist

The Demographics block reads xMap's population grid: hexagons of about 0.1 square kilometres, each with its residents, men and women, and nine age groups. Set to every hexagon, it returns the cells of a city one by one, which is exactly the unit a gap analysis needs. The age groups matter for transit: a block with many residents aged 60 and over, far from a stop, is a strong case for a demand-responsive minibus rather than a fixed route.

Your own stops, not someone else's list

xMap's places database has bus stops, tram stops and stations, and Nearby places can count them around each cell. For a first look that is enough. For a plan, use the agency's own stop list: upload it as a CSV and let Find nearest measure, for each populated hexagon, the distance to the closest of your stops (it reads up to 2,000 points with their positions). Your list knows about the stop moved last month and the temporary stop for the roadworks; a public map may not.

A straight line is not a walk

Find nearest measures straight-line distance. In a city with a river, a railway cutting or a motorway between a block and its stop, the real walk can be much longer. To check the gaps that look borderline, draw a walking area instead: Travel area takes each stop and draws the area reachable in, say, 10 minutes on foot along the streets. Run Demographics inside those areas and you get the residents who can actually walk to the network. Subtract that from the city's total and you have the people outside it, plain arithmetic on two numbers the flow gives you.

Use the walking areas for the shortlist, not the whole network: drawing an area for every stop of a large city is slow and unnecessary when the straight-line pass has already ruled most blocks in or out.

A worked example (made up)

Say a planning team in Paris wants to know where a new local bus loop would help most in the eastern districts. The figures below are invented to show the method; they are not measurements.

  1. Demographics, every hexagon, inside an Area of interest drawn around the eastern districts: say 2,400 populated cells come back.
  2. Sort and limit keeps the 300 most populated cells, since a loop must serve enough people to justify its cost.
  3. Find nearest, with the agency's stop list uploaded, adds the distance to the closest stop and the lines serving it.
  4. A Filter keeps cells whose nearest stop is more than 500 metres away: say 41 cells, home to about 60,000 people in this made-up case.
  5. Travel area draws a 10-minute walk from the nearest stop of each of those cells, to check which really are cut off and which only look far on a straight line.
  6. xMap LLM groups the remaining cells into two or three possible loops and writes a short note on each, which the team reviews before anything is proposed.

The output is not a decision. It is a list of places worth a site visit and a closer look at the timetable, produced in an afternoon instead of a fortnight.

Mistakes that skew the result

  • Counting stops that are not served at the times people travel. A stop with only a school-day service fills the coverage figure and helps nobody at seven in the evening. Join the frequency from your own data before you call a block covered.
  • Treating every resident the same. A block of mostly working-age adults near a rail station has other options; a block of older residents on a hill does not. Use the age groups.
  • Forgetting where people go. Residents are only half the picture. Hospitals, colleges and large employers are destinations; Nearby places can list them so the loop connects homes to somewhere useful.
  • Reading the grid as a census of today. A population grid is a snapshot, not a live count. It is good for comparing one block with another, less good for a new estate that filled up this year.

Keeping it current with an AI employee

Networks change: stops move, lines are cut, new housing opens. Rather than redoing the study once every few years, an AI employee in xMap Flows can rerun it each month and say what moved. The ready-made template Blocks with many people and few bus stops does the first version: residents in every block, the bus stops within 500 metres of each, the least served first, a short note for planners, a map and an email. New apartment buildings and malls that need a bus watches for new developments with no stop nearby, and On-demand minibus zones for older residents leans on the age groups.

What lands in the planning inbox is a ranked list of blocks with their residents and nearest stop, a map you can share with colleagues, and a few sentences that say which gaps are new since last month. The analysts keep the judgement; the flow keeps the data fresh.

For the wider method, see the Learn pages on white space analysis and drive time versus radius, and the public transport industry page.

Questions

What distance to a stop counts as well served?

Agencies set their own standard, often a walk of 400 to 800 metres depending on the service and the terrain. The method here works with any threshold: set the Filter to your agency's rule.

Can I use this in a city where population figures are not available?

Yes, but without residents. You can still find the housing estates, schools and hospitals far from a stop with Nearby places and Find nearest. The population weighting needs a country the grid covers.

Does the analysis take frequency into account?

Only if you add it. Put each stop's lines or departures per hour in your uploaded stop list, and Find nearest carries them onto each block so a Filter or the AI note can use them.

How often should it be rerun?

Monthly is enough for most networks, or after any timetable change. The flow can run on a schedule and email the team only when the list of gaps changes.

Templates that do this

Read next

More on workflows by industry · Every article · Templates

Put it to work

An AI employee in xMap Flows does this on live data, on its own timetable. Start from a template or describe it in plain words.